Suchergebnisse

One of the main goals of machine learning is to make a General Artificial Intelligence. Currently, human artificial intelligence researchers work on meticulously manipulating model parameters by hand in order to arrive at highly optimized machine learning models. In the future, a system will be needed such that a software is able to completely arrive at an optimized model to a specific topic all by itself. An increasingly aware human problem is stress, which can oftentimes lead to a variety of health issues. In this study, a novel machine learning platform was created that could learn how to assess the environmental factors relating to stress in a knowledge graph all by itself. Deep learning algorithms, in particular Graph Convolutional Algorithms, were employed to train the software on a small subset of topics in the aim of recreating additional knowledge graphs through automated Internet searches. By using constructed knowledge graphs with input plaintext for specific areas of environmental stressors as a dataset - weather, income, and societal class-, the software was able to accurately train and predict knowledge graphs for environmental stressors outside of its specific training domain for human analysis. These knowledge graphs could then be used in order to diagnose the total environmental stress through an analysis of how much a specific environment would traverse down the constructed knowledge graph.